fast-get: fix userspace

[lyakh]

Fix userspace z_vrfy_* implementations. Fixes a recent regression, caused by 879ba45 from #10379

[Copilot]

Pull request overview

This PR fixes a regression in the userspace syscall verification functions (z_vrfy_fast_put and z_vrfy_fast_get) that was introduced in commit 879ba45 from PR #10379. The fix addresses compilation/runtime issues in the CONFIG_USERSPACE code path.

• Corrected syscall marshalling include file names (removed incorrect 'mod_' prefix)
• Removed invalid memory validation calls attempting to access non-existent heap->heap.heap structure
• Properly organized includes for userspace syscall handlers

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[lyakh]

@lrudyX the PR only modifies code under #ifdef CONFIG_USERSPACE and AFAIK no QB tests are enabling that option, so the single failure of the HDA link DMA test in this PR cannot be related, please confirm

[lrudyX]

@lrudyX the PR only modifies code under #ifdef CONFIG_USERSPACE and AFAIK no QB tests are enabling that option, so the single failure of the HDA link DMA test in this PR cannot be related, please confirm

One test failed. Not related.

[lgirdwood]

Still no data on Jenkins, rerun.

[lgirdwood]

SOFCI TEST

[softwarecki]

Since this fixes a bug introduced due to the rushed review in #10379, I have to approve. However, I do not understand why these functions are turned into syscalls when the previous commit extended their parameter list with heap. It seems that for userspace modules fast_get should copy data to the heap accessible to the user, so the syscall is unnecessary. On the other hand, a syscall would make sense if we wanted to place the data in a read-only heap shared between modules, but then the heap parameter becomes redundant.

Regardless of that, please wrap all userspace-related changes in more specific ifdefs instead of the generic CONFIG_USERSPACE.